SYSTEM BY GNOSYS

loutclankedΤεχνίτη Νοημοσύνη και Ρομποτική

13 Νοε 2013 (πριν από 3 χρόνια και 11 μήνες)

66 εμφανίσεις

BUILDING A COGNITIVE
SYSTEM BY GNOSYS


Co
-
ordinator: John Taylor (KCL)


Asst Co
-
Ordinator: Stathis Kasderidis (FORTH)

EC PO: George Stork

Start date: Oct 1; Kick
-
off Oct 20/21


gnosys@ics.forth.gr

Web
-
site: http://www.ics.forth.gr/gnosys/


Department of Mathematics

King’s College London, UK

emails: john.g.taylor@kcl.ac.uk




CONTENTS

1.
Vision of GNOSYS

2.
GNOSYS Partners

3.
GNOSYS Prototypes

4.
GNOSYS Tasks/Milestones

5.
GNOSYS Summary


1. VISION OF GNOSYS

1)
Embodied cognition (wheel
-
robot + gripper)

2)
Create concepts/rewarded
-
goals under
attention control

3)
Learns goal
-
directed tasks

4)
Learns novel environments

5)
Reasoning by forward models

6)
Guidance from brain (animal/infant/adult)

7)
Various memory types
(STM/LTM/associative/error
-
based)

8)
Interdisciplinary: Comp vision/ Cog NSci/
Neural Networks/ Robotics/ AI/ Maths


General Work Plan of GNOSYS

GNOSYS Cognitive Powers


Feature
-
based perception (M1
-
16) WP2


Concepts/Goals/Attention (Sensory & Motor)


(M6
-
18, 12
-
24) WP2/WP3


Rewarded drive
-
based learning (M12
-
24) WP2


Goal
-
based Global Computation (M6
-
18) WP2


Abstraction Hierarchy (M12
-
24) WP3


Reasoning/Action Planning by motor
attention
-
base forward models(M18
-
33) WP3


Robot Platforms @ 2 levels (M18/M30)


HOW GNOSYS WORKS


v v


│ ↔ │


│ ↔ │


│ ↔ │


│ ↔ │


│ ↔ │



ANN

Adaptive


Stream

(Concepts/

Goals/

Attention/

Rewards/

Values/

Forward

Models

learnt as NN

predictors)


Symbolic


Control


Threads

(5 components)


Linguistic


Connections

(Words/Fuzzy rules/


Symbolisation)

: Relate to COSPAR

Drives/Motivation/Rewards


Assign values (in AMYG/OBFC) as direct
input (learnt), or by ‘DA’ modulation from
primary rewards (satisfying basic drives)


Basic drives for GNOSYS:


Energy level/ Curiosity/ Stimulation/
Minimum pain (touch/pressure)/
Approbation/ Motor activity


Use value maps
--
> assign value to stimuli



2. GNOSYS PARTNERS


1

King’s College London (KCL) Comp Nsci
Grp
: NNs, concepts, attn control

2
ZENON S.A., Greece (ZENON):

robots

3

Foundation of Research & Technology
-

Hellas Greece (FORTH):

global
comput/robots

4

Eberhard
-
Karls
-
Universität, Tübingen,
Germany (UTUB):

perception/reward/robots

5

Università di Genova, Dipartimento di
Informatica, Sistemistica, Telematica,

Italy (UGDIST):

motor control/robots



-
> RobotCub


Attentional Agent Architect
(EC FP5: DC, 2001
-
2003)


Distributed entity with four layers
(
attentional multi
-
level agent
):


L1: Sensors


L2: Pre
-
processing


L3: Local decision


L4: Global decision




GLOBAL CONTROL ARCHITECTURE


EXTENDED ATTENTION V EMOTION
ARCHITECTURE (EC ERMIS, NF, 2002
-
4;
BBSRC: 2004
-
7):


(extended Corbetta & Shulman, 2002)


Inhibitory

Interaction

through ACG:

Excitatory

interaction

Excitatory

Excitatory/Inhibitory

Inhibition from DLPFC

In emotion recognition

Endogenous

goals

Exogenous

goals

MOTOR CORTEX ACTION NETWORK (NT, MH, OM & JGT)

(in NetSim for sequence learning; tested in PDs: J NSci24:702 )

MOTOR CORTEX

GLOBUS PALLIDUS

EXTERNAL

GLOBUS PALLIDUS

INTERNAL

NUCLEUS RETICULARIS

THALMUS

SUBSTANTIA NIGRA

PARS COMPACTA

CENTROMEDIAN

PARAFISCULAR NUCLEUS

SUBSTANTIA NIGRA

PARS RETICULARIS

SUB
-
THALAMIC

NUCLEUS

GLUTAMATERGIC INPUT

GABAERGIC INPUT

DOPAMINERGIC INPUT

SIMILAR STRUCTURES MODEL OBFC, DLPFC, ACG AND VLPFC

FROM OTHER
CORTEX + OTHER
THALAMUS

FROM
CEREBELLUM

TO OTHER CORTEX

THALAMUS

STRIATUM

Cerebellar Structure

& Associated Regions: For Insertions,

by error
-
based learning (with teacher)


BK



PONS



GrC



GoC



PK



IO



DCN+



DCN
-


GrC

granule cells

GoC

golgi cells

BK


basket cells

PK


purkinje cells


DCN

deep cerebellar


nuclei (excit.


& inhib.)


IO


inferior olive

PONS

pontine nuclei

HIPP

hippocampal regions

PFC

pre
-
frontal cortex


inhibitory conns.


excitatory conns.

HIPP


PFC

HIPPOCAMPUS & AMYGDALA (in NetSim for sequence learning,
and x20 speed
-
up in SWS) (MH, NT & JGT): as teacher

EPSRC: Ventral & Dorsal Concept
Learning (
-
> GNOSYS)

Ventral pathway

V1

V2

V4

TEO

TE

LGN Input

Dorsal pathway

V1

V5

LIP

LGN Input

Learning

Hard
-
wired

Currently

Hard
-
wired

Architecture Details: Percepts


V1: 4 excitatory & inhibitory layers for bar
orientations, hardwired (14*14)




V2 (28*28) trained on reduced set of pairs of
bars (6), # start positions in retina 121




V4 (28*28)
-
>TEO (28*28/14*14)
-
>TE (7*7)
trained on 2 different triangles (121 start
positions)


Now by cluster computing


Next step: to DL/VLPFC as goals
-
> attention

ERMIS/BBSRC: GLOBAL BRAIN
CONTROL by ATTENTION:

Fusiform Gyrus

PFC

ACG/TPJ

PL

PL

VCX

PL

-
> Simulated Attentional Blink NF/JGT

-
> Consciousness by CODAM (Prog Neurobiology 03)

Model of Visuo
-
Motor Attention
Control System


(JGT + NF, IJCNN’03)

-
> MACS for

Attention filtering

-
>MINDRACES

for anticipation

AB extended by AMYG as bias:
ERPs for T2 in Lag3 when no
amygdala

ERPs for T2 in Lag3: amygdala input from T2

s
object rep, & fed back to same site



Tsuji T, Tanaka Y, Morasso P, Sanguineti V. Kaneko M (2002) IEEE Trans SMC
-
C, 32, 426
-
439.

Morasso P, Sanguineti V, Spada G (1997) Neurocomputing, 15, 411
-
434

UGDIST: Biomimetic trajectory formation

via artificial potential fields

… the importance of smoothness and continuity …


Khepera: the artificial body

The in
-
vitro brain

From the Neurobit project

Real
-
time control of robot motion

by sub
-
symbolic neural activity

… the importance of bidirectional communication …


… the importance of softness and a soft touch …

Robotized haptic interface

Comput
ational

Vision and Robotics

Lab (CVRL)


Institute of Computer Science


Foundation for Research
andTechnology


Hellas
(FORTH)

CVRL
-

FORTH


Mission:

Study the mechanisms involved
in the development of autonomous robotic
systems

Cognition

Action

Learning

Perception

System

Architecture

Right sub-network
Left sub-network
S.O.
n-1
A
B
M
A
B
C
M
C
Right sub-network
Left sub-network
A
B
M
A
B
C
M
C
Right sub-network
Left sub-network
A
B
M
A
B
C
M
C
BS-L
BS-R
S.O.
n
S.O.
n+1


Current

R&D

activities




p
erceptual

competences

based

on

visual

and

range

sensors

and

sensor

fusion

techniques


c
oupling

of

perception

and

action



autonomous

n
avigation

and

control

of

complex

robotic

systems


d
evelopment

of

networked

robotic

systems


content
-
based

retrieval

of

images

and

video


Future

activities


d
evelopment

of

robotic

behaviours

that

simulate

corresponding

behaviours

of

living

organisms


emergence

of

cognition

in

artificial

systems


c
omplex

heterogeneous

robotic

systems

involving

multiple

robots

CVRL
-

FORTH

UTUB


Experienced in robot movement
and planning

Involved in GNOSYS perceptions &
rewards


ZENON

Robotics Company in Athens

Experienced in robot applications

To construct robot platforms (2)

3. GNOSYS PROTOTYPES


PROTOTYPE I (M18): Attn control of sensory
inputs & response


Learn concepts of simple shapes [3] & rewarded
actions, under attention


Responses to commands/learn new goals as
new actions on new objects


PROTOTYPE 2 (M28): As above but more
complex objects [3] + sequences of action/object
pairs in real scenes + forward models for virtual
goal seeking (reasoning)

4. GNOSYS TASKS, etc:
Reasoning Domains/Environments
(WP2&3)


Three levels of environment


Level 1: Learn shapes/colours; move &
touch; move & pick up; [2] & [3]
-
D objects


Powers: Concept/Attn/Goals as actions on
objects/Valence of objects in environment


Level 2: Complex objects & actions


Powers: ibid/manipulate to achieve goals


Level 3: Hierarchy of objects; run virtual
object/action sequences to achieve goals


Powers: Reasoning/ novel objects/actions

Application to Patrolling, etc


Construct loc/action and object/action map
in patrol environment


Reasoning tasks: to discover actions: (loc1,
action)
→loc2, (obj1,action)→obj2


Meets barrier of boxes. Reasoning: move box
to pass through, instead of moving round
barrier


Over pond: reasoning: find plank to put
across pond


Plus many psychological tasks (WCST/Tower
of London, etc, etc)

MILESTONES


Level 1: Simple actions & stimuli [2] (M6)


Level 2: More complex actions & stimuli
[3]/colour/motion/audition/touch (M16)


Level3: Real
-
world stimuli (M24)


Prototype 1 (M18)


Prototype 2 (M28)


Assessment (M34)

5. GNOSYS SUMMARY


Create concepts/goals by learning


Can handle novel environments


Embodied cognitive system


Learning by infant
-
style development (by
hierarchy of modules sequentially coming
on line)


Reasoning by forward models created by
reward
-
based learning